J. Med. Microbiol. Ð Vol. 50 (2001), 49±54 # 2001 The Pathological Society of Great Britain and Ireland ISSN 0022-2615

HOST RESPONSE TO INFECTION

Reduced bactericidal activity against Staphylococcus aureus and Pseudomonas aeruginosa of blood neutrophils from patients with early adult respiratory distress syndrome M. T. MASCELLINO, G. DELOGU  , M. R. PELAIA, R. PONZO, R. PARRINELLO  and A. GIARDINA  Department of Infectious Diseases and  Institute of Anesthesia and Intensive Care, `La Sapienza' University, 00161 Rome, Italy

This study investigated the bactericidal capability of circulating neutrophils from blunt trauma patients admitted to an Intensive Care Unit against Staphylococcus aureus and Pseudomonas aeruginosa. Among those patients, two groups were considered and compared: patients who developed adult respiratory distress syndrome (ARDS) and patients who developed only pneumonia. Peripheral blood samples were drawn as soon as a diagnosis of pneumonia or ARDS was made, followed by the isolation of neutrophil cells and assessment of bacteria phagocytosis and killing. The results demonstrated that in patients with ARDS, phagocytosis and killing ef®ciency were signi®cantly impaired in comparison with patients with pneumonia and healthy controls. A possible dysregulation of reactive oxygen species production involving the release of humoral mediators in early ARDS may be involved.

Introduction It is well known that the non-speci®c immune response to bacterial infection is mediated primarily by neutrophil polymorphonuclear leucocytes (PMNLs), which represent the `professional' phagocytic cells deployed to eradicate invading micro-organisms. PMNLs are also involved in the pathogenesis of many in¯ammatory syndromes including the adult respiratory distress syndrome (ARDS) [1, 2]. Despite many advances in supportive care, ARDS still has a high mortality rate and there is clear evidence that ARDS patients usually die of infection, particularly lower respiratory tract infections [3]. Patients with ARDS have been shown to have alterations in the function of circulating neutrophils or those in the lungs, or both, and such a dysfunction could contribute to the enhanced risk of infection in these patients [4, 5]. The present study sought to evaluate the bactericidal ef®ciency of circulating neutrophils from patients admitted to an Intensive Care Unit (ICU). Patients with blunt chest trauma who developed pneumonia or Received 22 March 2000; revised version received 28 May 2000; accepted 6 June 2000. Corresponding author: Dr M.T. Mascellino.

ARDS were compared, as soon as a diagnosis of pneumonia or ARDS had been made.

Materials and methods Study design The study was performed after approval by the institutional ethics committee and informed consent was obtained from each patient or their next of kin. Sixty subjects admitted to an Intensive Care Unit (ICU) with blunt chest trauma were enrolled over a 28-month period (Jan. 1998±Jan. 2000). Patients were excluded from the study for the following reasons: need for antibiotic therapy; presence of infection on admission; age ,18 years; suspected pregnancy or post-partum state; immunosuppressed state (i.e., treatment with steroids, bones marrow or organ transplant recipients, haematological malignancy and AIDS); medical condition considered to be irreversible. The acute physiology and chronic health evaluation (APACHE II) score was employed to determine the initial severity of illness [6]. Two groups of patient were investigated: those who developed pneumonia during hospitalisation and those who developed ARDS. The following criteria for clinical diagnosis of

50

M.T. MASCELLINO ET AL.

pneumonia were used: radiographic evidence of a new and persistent pulmonary in®ltrate (other than those of non-infectious origin), fever (>388C), leucocytosis ($ 12 3 109 =L) and purulent respiratory secretions. ARDS was de®ned by the presence of classic criteria revised by the American-European Consensus Conference [7] as follows: a compatible underlying disease, a pulmonary artery occlusion pressure of ,18 mmGH during mechanical ventilation and bilateral lung opacities. On the diagnosis of pneumonia or ARDS, blood samples were withdrawn to isolate PMNLs and to measure adhesivity, phagocytosis and killing of Pseudomonas aeruginosa (ATCC 27853) and Staphylococcus aureus (ATCC 6538). These organisms were chosen because they represented the species most frequently isolated from broncho-alveolar lavages (BALs) of patients admitted to the ICU. Ten healthy controls were selected and compared with the population under study. These controls were matched by age, sex and seasonal time of enrolment.

Preparation of human PMNLs PMNLs were collected from healthy donors and from the patients by Boyum's technique [8]. PMNLs were used within 1 h of isolation. Cell preparations were .95% neutrophils by Diff-Quick staining (Baxter Scienti®c Products, Miami, FL, USA) and were .96% viable in trypan blue exclusion tests.

Labelling and opsonisation Attachment and ingestion of bacteria by PMNLs were assessed by direct visualisation with a modi®cation of a previously described ¯uorescence microscopy method [9]. For each phagocytosis assay, an overnight culture of micro-organisms in Brain Heart Infusion (BHI; Merck Darmstad Germany) broth was diluted 1 in 100 in fresh BHI broth and grown to mid-log phase. Bacterial density was adjusted spectrophotometrically to a concentration of 2 3 108 cfu=ml. Bacteria were labelled by incubation with ¯uorescein isothiocyanate (FITC; Sigma) 0.1% in 50 mM sodium carbonate buffer, pH 9.6, for 30 min at 378C in the dark. Labelled organisms were washed twice and suspended in Hanks's Balanced Salts Solution (HBSS) without Ca2‡ and Mg2‡ . The FITC-labelled bacteria were then opsonised by incubation with pooled normal human serum 10% in HBSS at 378C for 15 min with rotation. After opsonisation, bacterial cells were pelleted, washed and resuspended in HBSS.

Phagocytosis Each phagocytic mixture contained 0.5 ml of opsonised ¯uorescein-labelled bacteria (2 3 107 cfu=ml) and 0.5 ml of PMNLs (2 3 106 =ml) in HBSS plus 1 mM Ca2‡ and 1 mM Mg2‡ . The ratio of bacteria to cells was 10 to 1. The tube was rotated for 30 min at 378C, then 200-ml samples were removed. In experiments

with S. aureus, the mixtures were treated with lysostaphin 10 mg=l for 5 min at 48C to kill any residual extracellular bacteria. Incubation at 48C prevented any effect of lysostaphin on intracellular S. aureus, as lysostaphin does not enter cells at low temperatures but does eliminate extracellular bacteria. Ten ml of ethidium bromide 0.1% were added to each phagocytic mixture to a ®nal concentration of 50 ìg=ml. Then 10 ìl of the mixture were placed on a glass slide, overlaid with a coverslip and examined, within 20 min, with a ¯uorescence microscope with a 520-nm FITC ®lter under oil immersion (magni®cation, 31000). Surface-attached or extracellular bacteria appeared orange, whereas ingested micro-organisms showed a rim of intense apple-green ¯uorescence without any orange staining. At least 20 PMNLs per sample were examined to evaluate the number of attached or ingested organism per cell [10]. Intracellular viable bacteria were determined by standard colony counts (cfu/ml).

Killing Bacteria suspended in HBSS and PMNLs at a ratio of 10:1 were incubated in a water bath with agitation at 378C (in polypropylene vials) for 3 h. Bacteria without PMNLs were also checked to determine bacterial growth and to prevent spontaneous agglutination during the incubation time. At 60, 120 and 180 min samples were taken in triplicate, mixed with distilled water to lyse the PMNLs and, after 10-fold serial dilutions, suspended in molten BHI agar for enumeration of cfu/ml after 24 h (determination of the number of intracellular viable bacteria) Bacterial multiplication in the medium over a 30-min period was eliminated as a factor by counting the bacterial inoculum incubated in the medium in the absence of PMNLs: no signi®cant increase occurred during the 30-min period.

Assessment of bacterial phagocytosis, killing and association to PMNLs Phagocytosis was expressed as the decrease in the initial number of viable extracellular bacteria in the inoculum by the following formula: Ph(t) ˆ N0 N1 in which Ph = Phagocytosis at time t, N0 is the number of viable bacteria incubated in medium for 30 min at 378C in the absence of PMNLs and N1 is the number of viable extracellular bacteria in the presence of PMNLs after incubation for 30 min. Bacterial killing was calculated by the following equation: K(t) ˆ Ph(t) VB(t) in which K (t) is killing at time t, Ph (t) is phagocytosis at time t and VB is the number of viable intracellular bacteria at time t [11]. Bacterial association to PMNLs was evaluated by examining at least 200 PMNLs and determining the percentage of cells which had two or more associated bacteria.

NEUTROPHIL BACTERICIDAL DYSFUNCTION AND ARDS

51

Results

Detection of PMNL receptors Monoclonal antibodies (MAbs) (Becton-Dickinson, Milan, Italy) against the PMNL receptors CD11 and CD16 were used. The former is speci®cally directed towards the complement fraction C3b, whereas the latter is speci®c for the Fc component of the immunoglobulins. Heparinised blood (20-ml amounts) from the patients was added to two different tubes with 5 ml of MAbs ± anti-CD11 conjugated with ¯uorescein (FITC) and anti-CD16 conjugated with phycoerythrin (PE), respectively. Controls were run with blood from healthy donors. The tubes were incubated for 20 min in the dark, then 1 ml of a lysis solution was added to destroy the red cells. After further incubation for 10 min in the dark, the suspensions were centrifuged for 5 min at 380 g. The pellets were washed with 1 ml of PBS and centrifuged for 5 min at 380 g. After removing the supernates, the pellets were suspended again in 250 ml of PBS. Flow cytometric readings were taken after careful mixing of the suspensions. Each suspension was aspirated and then passed through a laser beam emitting either red (for CD16) or green (for CD11) ¯uorescence, the intensity of which corresponded to the expression of each receptor. The results were reported as the percentage of PMNLs expressing the two receptors (semi-quantitative results).

Statistical analysis Data are expressed as mean and SD. Wilcoxon's rank sum test was used for comparison of values between pneumonia patients and ARDS patients, with a signi®cance threshold of p , 0.05. The Mann-Whitney test was employed to compare the patient characteristics.

Patients Of the 60 patients enrolled at the beginning of the study, 38 patients were followed: 20 who developed pneumonia during hospitalisation and 18 who developed ARDS. Table 1 shows the demographic data and the APACHE II scores, as well as the mortality rate of pneumonia patients and ARDS patients. There was no signi®cant difference between the two groups of subjects. Mechanical ventilation was performed with a volume-controlled ventilator (Servo 9000 C; Siemens Elema Solna, Sweden) and continuous analgesic sedation with midazolam and fentanyl was maintained. Ventilator settings, level of positive end-expiratory pressure and fractional inspired oxygen (F102 ) were adjusted as necessary. All patients were treated with supportive ¯uid infusion, histamine-2-inhibitor and, as needed, cardiac glycosides. The mean hospital stay was 12 (SD3) days.

Bacterial association with human PMNLs Table 2 shows the mean ingestion of P. aeruginosa and S. aureus by PMNLs from pneumonia patients and ARDS patients compared with PMNLs from healthy controls. No signi®cant difference in percentage uptake was found either in the three groups of patients or in the different bacterial targets: in fact, a mean value of ingestion equal to 36.6% (range 28±45) was observed for the three groups of patients.

Phagocytosis and killing A correlation of 0.9 between cfu and ¯uorescence values established the validity of both microbiological and microscopic assays in the assessment of intracellu-

Table 1. Demographic data, APACHE II score and mortality rate of patients with pneumonia or ARDS Group

n

Age

Sex (M/F)

APACHE II score

Mortality rate (%)

Pneumonia patients ARDS patients

20 18

59 (12) 59 (17)

11/9 7/11

19 (2) 23 (3)

29 43

 Data are expressed as mean (SD).

Table 2. Ingestion of P. aeruginosa and S. aureus by PMNLs from each patient group. Uptake(%)

Source of PMNLs

n

P. aeruginosa

S. aureus

Healthy controls Pneumonia patients ARDS patients

10 20 18

45.0 (12.5) 40.1 (9.8) 38.2 (7.9)

36.2 (9.7) 32.5 (8.9) 28.0 (8.7)

Data are expressed as mean (SD).

52

M.T. MASCELLINO ET AL.

lar viable bacteria. The results obtained with S. aureus in the study population are given in Fig. 1. During a 30-min incubation of 2:0 3 107 cfu of bacteria/ml with PMNLs (2:0 3 106 ), the parameters of phagocytosis and killing were calculated by using the cfu and the ¯uorescence assay. When the numbers of S. aureus phagocytosed and killed in healthy controls were used as base-line values, valid comparisons could be made; the numbers of S. aureus phagocytosed and killed by healthy controls were calculated as 3.75 (SD 0.34) 3 106 and 3.25 (SD 0.30) 3 106 cfu=ml, respectively. In ARDS patients, S. aureus was less ef®ciently phagocytosed and killed during the same 30-min, period ± 1.50 (SD 0.24) 3 106 and 1.25 (SD 0.20) 3 106 cfu=ml. The decrease in phagocytosis and killing in this group of patients was statistically signi®cant compared with healthy controls and pneumonia patients (p , 0.05), in which the bacteria phagocytosed and killed were 3 (SD 0.25) 3 106 and 2.75 (SD 0.20) 3 106 cfu=ml, respectively. The study also demonstrated that S. aureus was more readily eliminated in the controls than in the other two groups. The proportion of killed to surviving bacteria phagocytosed by PMNLs during a 30-min period was higher in the controls (5.0) than in pneumonia patients (3.05) or ARDS patients (1.95) (data not shown). The data obtained for P. aeruginosa were similar to those obtained for S. aureus (Fig. 2). Again, it was evident that the difference in phagocytosis and killing between the ARDS patients and healthy controls was statistically signi®cant (p , 0.05), whereas no signi®cant difference was observed between the controls and pneumonia patients. P. aeruginosa appeared to be less amenable to phagocytosis and intracellular killing than

S. aureus. In fact, the numbers of phagocytosed and killed P. aeruginosa were lower than the correspondent values of S. aureus: 1.5 (SD 0.24) 3 106 and 1.25 (SD 0.20) 3 106 cfu=ml in healthy controls; 8 (SD 0.25) 3 105 and 6 (SD 0.18) 3 105 cfu=ml for pneumonia patients; 3.20 (SD 0.8) 3 104 and 3 (SD 0.5) 3 104 cfu=ml for ARDS patients, respectively.

Analysis of PMNL receptors The measurement of PMNL receptors is reported in Table 3, which shows the mean values for both groups of patients (with ARDS and with pneumonia) and the controls. No difference was evident among the three groups; in fact, the percentages of CD11 and CD16 were c. 99% and 1.5%, respectively, in all groups.

Discussion The data from the present study indicated that in blunt trauma patients, those who developed ARDS early exhibited an alteration of circulating PMNLs in terms of both reduced phagocytosis and decreased killing of bacterial strains such as S. aureus and P. aeruginosa. This defect appears signi®cantly obvious if compared with the results obtained in blunt trauma patients who developed only pneumonia. By contrast, no differences in bacterial ingestion by PMNLs was found in the two groups of patients nor in the controls. This observation was sustained by the ®nding of a similar percentage of PMNL receptors in the population under study. Indeed, it was observed that

4.5⫻106

Number of S. aureus (cfu/ml)

3.5⫻106

2.5⫻106

* 1.5⫻106

5⫻105

I

II

III

Fig. 1. S. aureus (cfu/ml) phagocytosed (h) and killed ( ) by PMNLs from healthy controls (I), pneumonia patients (II) and ARDS patients (III). Data are expressed as mean and SD.  p , 0.05 versus results for controls and pneumonia patients.

NEUTROPHIL BACTERICIDAL DYSFUNCTION AND ARDS

53

1.8⫻106

Number of P. aeruginosa (cfu/ml)

1.5⫻106

1.2⫻106

9⫻105

6⫻105

3⫻105

* 5⫻104 0

I

II

III

Fig. 2. P. aeruginosa (cfu/ml) phagocytosed (h) and killed ( ) by PMNLs from healthy controls (I), pneumonia patients (II) and ARDS patients (III). Data are expressed as mean and SD.  p , 0.05 versus results for controls and pneumonia patients. Table 3. Percentage of PMNL receptors (CD11, CD16) in the study population (semi-quantitative results) PMNL receptors (%) Group

n

CD11

CD16

Healthy controls Pneumonia patients ARDS patients

10 20 18

99.2 (12.3) 99.16 (9.8) 99.02 (11.5)

1.56 (0.8) 1.3 (0.6) 1.74 (0.5)

Data are expressed as mean (SD).

PMNLs from all groups expressed similar levels of receptors such as CD11 and CD16, with c. 99% and 1.5% for CD11 and CD16, respectively. It is conceivable that the reduced bactericidal capacity of the blood PMNLs found in ARDS patients could be attributable to an altered cell function, namely an impairment of respiratory burst activity, because of the central role played by the production of reactive oxygen species (ROS) [12]. Previous studies have shown PMNLs from ARDS patients show functional and metabolic alterations [13, 14], but the literature is contradictory regarding ROS generation. Some authors described an enhanced production while others found decreased release of superoxide anion (O2 ) and hydrogen peroxide (H2 O2 ). For example, Rivkind and co-workers observed activation of the neutrophil oxidative burst in circulating PMNLs from ARDS patients as assessed by superoxide anion secretion [15]. However, Parson et al. reported that PMNLs from patients with ARDS produced less superoxide than did cells from normal subjects when primed with lipopolysaccharide (LPS)

or stimulated with N-formyl-methionyl-leucine-phenylalanine [16]. Chollet-Martin et al. found that circulating PMNLs from ARDS patients exhibited a biomodal response of H2 O2 production in ¯ow cytometry studies [17]. This observation con®rmed the functional heterogeneity of PMNLs described elsewhere [18] and could explain in part the discordant results obtained so far. An impairment of both bactericidal and oxidative burst characteristics was noticed not only in circulating but also in alveolar cells isolated from lungs by broncho-alveolar lavage (BAL) [19]. CholletMartin et al. showed dysregulation of activity, and this could explain, in the opinion of the authors, the increased susceptibility to bacterial infections in ARDS patients [20]. In line with this ®nding, other studies demonstrated a reduced phagocytosis of Escherichia coli by alveolar macrophages during endotoxaemia and a down-regulation of ROS production by the same macrophages was noted during P. aeruginosa bacteraemia in an experimental animal model [21, 22].

54

M.T. MASCELLINO ET AL.

Unfortunately, in the present study the production of reactive oxygen metabolites was not measured. Nevertheless, on the basis of the data reported, the results seem to support indirectly the theory of a downregulation of the ROS secretion by circulating PMNLs in early ARDS. The exact mechanism causing this defect is still not clearly understood; some studies point out the main involvement of humoral mediators and the interaction between PMNLs and cytokines [23, 24]. High levels of tumour necrosis factor-á (TNF-á), as well as of interleukin-6 (IL-6) and IL-8, have been found in plasma and in BAL of ARDS patients [25, 26]. These compounds are able to activate PMNL function in vivo, accounting for the correlation observed between their enhanced systemic concentration and the expression of CD11b and CD62L receptors [20]. Yuan et al. observed in vitro the effects of some proin¯ammatory cytokines including granulocyte-macrophage colony-stimulating factors (GM-CSF), IL-2, IL6, IL-1á, IL-1â and IFN-â on intracellular oxidative production in normal PMNLs [27]. Whereas GM-CSF had a marked in¯uence on H2 O2 release, other mediators such as IL-1á were less effective. On the contrary, IL-10, an immunomodulatory cytokine which shows an anti-in¯ammatory effect, would inhibit the ROS generation by PMNLs [28]. In early ARDS, pathways could be triggered inducing the release of humoral mediators which down-regulate the PMNL bacterial activity by interfering with ROS generation. However, to test and con®rm this hypothesis, further studies, addressing the interaction between systemic in¯ammatory mediators and the bactericidal function of PMNLs in ARDS, will be required.

References 1. Lehrer RI, Ganz T, Selsted ME, Babior BM, Curnette JT. Neutrophils and host defense. Ann Intern Med 1988; 109: 127±142. 2. Swank DW, Moore SB. Roles of the neutrophil and other mediators in adult respiratory distress syndrome. Mayo Clin Proc 1989; 64: 1118±1132. 3. Montgomery AB, Stager MA, Carrico CJ, Hudson LD. Causes of mortality in patients with the adult respiratory distress syndrome. Am Rev Respir Dis 1985; 132: 485±489. 4. Martin TR, Pistorese BP, Hudson LD, Maunder RJ. The function of lung and blood neutrophils in patients with adult respiratory distress syndrome. Implications for the pathogenesis of lung infection. Am Rev Respir Dis 1991; 144: 254±262. 5. Seidenfeld JJ, Pohl DF, Bell RC, Harris GD, Johanson WG. Incidence, site, and outcome of infections in patients with the adult respiratory distess syndrome. Am Rev Respir Dis 1986; 134: 12±16. 6. Chu DY. Predicting survival in AIDS patients with respiratory failure. Application of the APACHE II scoring system. Crit Care Clin 1993; 9: 89±105. 7. Bernard GR, Artigas A, Brighan KL, et al. The AmericanEuropean consensus on ARDS. De®nitions, mechanisms, relevant outcomes and clinical trial coordination. Am J Respir Crit Care Med 1994; 149: 818±824. 8. Boyum A. Isolation of lymphocytes, granulocytes and macro-

phages. Scand J Immunol 1976; Suppl 5: 9±15. 9. Ardunio RC, Jacques-Palaz K, Murray BE, Rakita RM. Resistance of Enterococcus faecium to neutrophil-mediated phagocytosis. Infect Immun 1994; 62: 5587±5594. 10. Drevets DA, Campbell PA. Macrophage phagocytosis: use of ¯uorescence microscopy to distinguish between extracellular and intracellular bacteria. J Immunol Methods 1991; 142: 31±38. 11. Adinol® LE, Bonventre PF. Enhanced phagocytosis, killing, and serum sensivity of Escherichia coli and Staphylococcus aureus treated with sub-MICs of imipenem. Antimicrob Agents Chemother 1998; 32: 1012±1018. 12. Henson PM, Johnston RB. Tissue injury in in¯ammation. Oxidants, proteinases, and cationic proteins. J Clin Invest 1987; 79: 669±674. 13. Laurent T, Markert M, Von Fliendner V et al. CD11b/CD18 expression, adherence, and chemotaxis of granulocytes in adult respiratory distress syndrome. Am J Respir Crit Care Med 1994; 149: 1534±1538. 14. Chollet-Martin S, Gatecel C, Kermarrec N, Gougerot-Pocidalo MA, Payen DM. Alveolar neutrophil functions and cytokine levels in patients with the adult respiratory distress syndrome during nitric oxide inhalation. Am J Respir Crit Care Med 1996; 153: 985±990. 15. Rivkind AI, Siegel JH, Littleton M et al. Neutrophil oxidative burst activation and the pattern of respiratory physiologic abnormalities in the fulminant post-traumatic adult respiratory distress syndrome. Circ Shock 1990; 33: 48±62. 16. Parsons PE, Gillespie MMK, Moore EE, Moore FA, Worthen GS. Neutrophil response to endotoxin in the adult respiratory distress syndrome: Role of CD14. Am Respir Cell Mol Bio 1995; 13: 152±160. 17. Chollet-Martin S, Montravers P, Gilbert C et al. Subpopulation of hyperresponsive polymorphonuclear neutrophils in patients with adult respiratory distress syndrome. Role of cytokine production. Am Rev Respir Dis 1992; 146: 990±996. 18. Bass DA, Ozbrantz P, Szejda P, Seeds MC, McCall CE. Subpopulations of neutrophils with increased oxidative product formation in blood of patients with infection. J Immunol 1986; 136: 860±866. 19. Jean D, Rezaiguia-Delclaux S, Delacourt C et al. Protective effect of endotoxin instillation on subsequent bacteria-induced acute lung injury in rats. Am J Respir Crit Care Med 1998; 158: 1702±1708. 20. Chollet-Martin S, Jourdain B, Gilbert C, Elbim C, Chastre J, Gougerot-Pocidalo MA. Interactions between neutrophils and cytokines in blood and alveolar spaces during ARDS. Am J Respir Crit Care Med 1996; 154: 594±601. 21. Jacobs RF, Kiel DP, Balk RA. Alveolar macrophage function in a canine model of endotoxin-induced lung injury. Am Rev Respir Dis 1986; 134: 745±751. 22. Jenkins JK, Carey PD, Byrne K, Sugarman HJ, Fowler AA. Sepsis-induced lung injury and the effects of ibuprofen pretreatment. Analysis of early alveolar events via repetitive bronchoaveolar lavage. Am Rev Respir Dis 1991; 143: 155±161. 23. Strieter RM, Kunkel SL. Acute lung injury: the role of cytokines in the elicitation of neutrophils. J Invest Med 1994; 42: 640±651. 24. Marie C, Muret J, Fitting C, Losser M-R, Payen D, Cavaillon J-M. Reduced ex vivo interleukin-8 production by neutrophils in septic and nonseptic Systemic In¯ammatory Response Syndrome. Blood 1998; 91: 3439±3446. 25. Millar AB Singer M, Meager A, Foley NM, McJohnson NM, Rook GAW. Tumour necrosis factor in bronchopulmonary secretions of patients with adult respiratory distress syndrome. Lancet 1989; 2: 712±714. 26. Miller EJ, Cohen AB, Nagao S et al. Elevated levels of NAP1/interleukin-8 are present in the airspaces of patients with the adult respiratory distress syndrome and are associated with increased mortality. Am Rev Respir Dis 1992; 146: 427±432. 27. Yuan L, Inone S, Saito Y, Nakajima O. An evaluation of the effects of cytokines on intracellular oxidative production in normal neutrophils by ¯ow cytometry. Exp Cell Res 1993; 209: 375±381. 28. Dandona P, Mohanty P, Hamouda W, Aljada A, Kumbkarni Y, Garg R. Effect of dexamethasone on reactive oxygen species generation by leukocytes and plasma interleukin-10 concentrations: a pharmacodynamic study. Clin Pharmacol Ther 1999; 66: 58±65.